Abstract

A large data set of ground-velocity time histories from earthquakes that occurred in Friuli-Venezia Giulia (northeastern Italy) was used to define regional predictive relationships for ground motion, in the 0.25- to 14.0-Hz frequency band. The bulk of the data set was provided by the seismic network run by Centro Ricerche Sismologiche (CRS), a department of the Istituto Nazionale di Oceanografia e Geofisica (OGS). A collection of 17,238 selected recordings from 1753 earthquakes was compiled for the years 1995–1998, with magnitudes ranging from Mw ∼1 to 5.6. Ninety-six three-component strong-motion waveforms belonging to the largest events of the 1976–1977 Friuli seismic sequence were also taken from the enea-enel accelerogram database and included in our data set. For the strongest event, which occurred on 6 May 1976 at 20:00 local time, an average local magnitude ML 6.6 was computed by Bonamassa and Rovelli (1986). The inclusion of a large number of acceleration time histories from this earthquake and six others, from magnitudes from Mw 5.2 to magnitude Ms 6.1 (three of them of Ms ∼6.0), extends the validity of the predictive relationships proposed in this study up to the highest magnitude ever recorded in the region.

A total of 10,256 vertical-component and 6982 horizontal-component seismograms were simultaneously regressed for excitation and site characteristics, as well as for the crustal propagation, in the hypocentral distance range 20–200 km. Results are given in terms of excitation, attenuation, and specific site for the vertical ground motion, together with a horizontal-to-vertical ratio for each existing horizontal-component seismometer. The regional propagation was modeled in the 0.5- to 14.0-Hz frequency band by using a frequency-dependent piece wise continuous linear (in a log–log space) geometrical spreading function and a frequency-dependent attenuation parameter:  
\[Q(f)=260(f{/}1.0)^{0.55}\]
The excitation spectra of larger events were modeled by using the regional propagation, a single-corner frequency Brune spectral model characterized by an effective stress parameter,  
\[{\Delta}{\sigma}=60{\ }\mathrm{MPa},\]
and by a regional estimate of the near-surface, distance-independent, network-averaged attenuation parameter  
\[{\kappa}_{0}=0.045{\ }\mathrm{sec}\]
that was estimated from the rolloff of the empirical source spectra obtained from the regressions. Other studies (De Natale et al., 1987; Cocco and Rovelli, 1989; Singh et al., 2001) suggested large stress drops (Δσ ≃ 30–100 MPa,) to explain the high-frequency amplitude levels of the seismic radiation of the largest quakes of the 1976 sequence.

Predictions for peak ground acceleration (PGA) and pseudo–spectral velocity (PSV) (5% damping) were computed through the use of the random vibration theory (RVT), with the parameters obtained from the regressions of this study.

You do not currently have access to this article.